pypomp.models.UKMeasles¶
- class pypomp.models.UKMeasles[source]¶
Bases:
objectClass interface for building UK Measles POMP models and accessing data.
This class constructs POMP models based on the He et al. 2010 [1] formulation and accesses epidemiological case reports data from Korevaar et al. 2020 [2].
Model Variants¶
The following sub-models (selected via the
modelparameter inPomp()andPanelPomp()) are available:“001”: Standard He10 model (He et al. 2010 [1]) including mixing parameter
alphaand death ratemu.“001b” (Default): He10 model without
alpha(mixing parameter fixed to 1) and withoutmu(death rate parameter omitted).“001d”: He10 model without
alphaormu(similar to “001b”), but configured with gradient-stable transitions and measurement densities for DPOP training.“002”: He10 model without
alphaormu, where the transmission import rateiotascales with population size:iota = exp(iota1 + iota2 * log(pop)). Designed for shared parameter panel modeling.“002d”: DPOP-enabled version of “002” combining the population-scaling
iotaparameterization with gradient-stable transitions/measurement densities.“003”: Variant of the He10 model formulated with a continuous process model.
Model Parameters
- R0float
Basic reproduction number.
Active Models: All models.
Estimation Scale: Log transformed.
- sigmafloat
Rate of transition from susceptible to exposed (1/latent period).
Active Models: All models.
Estimation Scale: Log transformed.
- gammafloat
Rate of transition from exposed to infectious (1/infectious period).
Active Models: All models.
Estimation Scale: Log transformed.
- iotafloat
Daily imported cases rate.
Active Models: “001”, “001b”, “001d”, “003” (omitted in “002”, “002d”).
Estimation Scale: Log transformed.
- iota1float
Baseline imported cases (used in population-scaling import rate formulation).
Active Models: “002”, “002d” (replaces
iota).Estimation Scale: Untransformed (natural scale).
- iota2float
Rate at which imported cases increase with log population size.
Active Models: “002”, “002d” (replaces
iota).Estimation Scale: Untransformed (natural scale).
- sigmaSEfloat
Rate of stochastic extrademographic variation / process noise.
Active Models: All models.
Estimation Scale: Log transformed.
- cohortfloat
Cohort effect coefficient reflecting entry rate of children into school.
Active Models: All models.
Estimation Scale: Logit transformed.
- amplitudefloat
Seasonality amplitude coefficient of transmission rate.
Active Models: All models.
Estimation Scale: Logit transformed.
- rhofloat
Reporting probability.
Active Models: All models.
Estimation Scale: Logit transformed.
- psifloat
Reporting error over-dispersion.
Active Models: All models.
Estimation Scale: Log transformed.
- mufloat
Background death rate.
Active Models: “001” only.
Estimation Scale: Log transformed.
- alphafloat
Mixing parameter/exponent for force of infection (e.g. transmission rate scales with
(I + iota)**alpha).Active Models: “001” only.
Estimation Scale: Log transformed.
- S_0, E_0, I_0, R_0float
Initial proportions of susceptible, exposed, infectious, and recovered populations.
Active Models: All models.
Estimation Scale: Transformed as log ratio proportions (normalized relative to sum).
References
Methods
AK_mles()Returns a data frame of Aaron King's MLEs from https://kingaa.github.io/sbied/measles/index.html
PanelPomp(units, theta[, model, ...])Returns a PanelPomp object for the UK Measles data.
Pomp(unit, theta[, model, interp_method, ...])Returns a Pomp object for the UK Measles data.
subset([units, clean])Return a subset of the UKMeasles data, filtered by the given units.
units()List of all unique unit names in the UKMeasles dataset.